Linköping University Medical Dissertations No. 1019
Linköping 2007
SURVEILLANCE OF
ANTIBIOTIC CONSUMPTION
AND ANTIBIOTIC RESISTANCE
IN SWEDISH INTENSIVE CARE UNITS
SURVEILLANCE OF
ANTIBIOTIC CONSUMPTION
AND ANTIBIOTIC RESISTANCE
IN SWEDISH INTENSIVE CARE UNITS
MARCUS ERLANDSSON
Linköping University Medical Dissertations No. 1019
Linköping 2007
Surveillance of Antibiotic Consumption and Antibiotic Resistance in Swedish Intensive Care Units © Marcus Erlandsson 2007
All previously published papers, figures and tables are reprinted with permission from the publishers. Date of disputation: 2007-10-26
Institution: IKE
ISBN: 978-91-85895-77-9 ISSN: 0345-0082
Art direction: Niklas Ramviken, ENO form Printed by LiU-Tryck, Linköping, Sweden, 2007
INTRODUCTION: Nosocomial
infec-tions remain a major cause of mortal-ity and morbidmortal-ity. The problem is most apparent in intensive care units (ICUs). Most ICU patients are compromised and vulnerable as a result of disease or se-vere trauma. One in ten people admitted to hospital is given an antibiotic for in-fection. The risk of acquiring a nosoco-mial infection in a European ICU is ap-proximately 20%. It is vitally important that ways are found to prevent transmis-sion between patients and personnel, and that local hygiene routines and antibiotic policies are developed. This thesis is a holistic work focused particularly on an-timicrobial antibiotic resistance, antibi-otic consumption and to some extent on hygiene in Swedish ICUs.
AIMS: The general aim of this thesis was
to investigate bacterial resistance and antibiotic consumption in Swedish ICUs and to try to correlate ICU demographic data with antibiotic consumption and an-tibiotic resistance. Additional aims were to investigate on which clinical indica-tions antibacterial drugs are prescribed
in the ICU, and to investigate the emer-gence of resistance and transmission of Pseudomonas aeruginosa in the ICU using cluster analysis based on antibio-grams and genotype data obtained by AFLP.
MATERIAL AND METHODS: In
pa-per 1-3, antibiotic consumption data together with bacterial antibiotic resist-ance data and specific ICU-demographic data were collected from an increasing number of ICUs over the years 1997-2001. Data from ICUs covering up to six million out of Sweden’s nine million inhabitants were included. In paper 4, the indications for antibiotic prescribing were studied during two weeks in 2000. Paper 5 investigated Pseudomonas aeru-ginosa isolates in order to detect cross-transmission with genotype obtained by AFLP, and antibiogram-based cluster analysis was also performed in order to see if this could be a quicker and easier substitute for AFLP.
RESULTS: This thesis has produced
three important findings. Firstly, antibi-otic consumption in participating ICUs was relatively high during the study pe-riod, and every patient received on aver-age more than one antimicrobial drug per day (I-IV). Secondly, levels of antimi-crobial drug resistance seen in S. aureus,
E. coli and Klebsiella spp remained low
when data were pooled from all ICUs throughout the study period, despite relatively high antibiotic consumption (I-V). Thirdly, the prevalence of antibi-otic resistance in CoNS and E. faecium, cefotaxime resistance in Enterobacter, and ciprofloxacin and imipenem resist-ance in P. aeruginosa was high enough to cause concern.
CONCLUSION: For the period studied,
multidrug resistance in Swedish ICUs was not a major problem. Signs of cross-transmission with non-multiresistant bacteria were observed, indicating a hygiene problem and identifying sim-ple improvements that could be made in patient care guidelines and barrier pre-cautions. A need for better follow up of prescribed antibiotics was evident. With further surveillance studies and moni-toring of antibiotics and bacterial resist-ance patterns in the local setting as well as on a national and international level, some of the strategic goals in the pre-vention and control of the emergence of antimicrobial-resistant microbes may be achievable.
ABSTRACT ... 6
TABLE OF CONTENTS ... 8
LIST OF PUBLICATIONS ... 10
ABBREVIATIONS ... 12
INTRODUCTION ... 14
Nosocomial ICU infections ... 15
ICU pathogens ... 16
Antimicrobial drugs ... 18
Susceptibility breakpoints ... 21
ATC, DDD and DDD1000 ... 21
Antimicrobial drug resistance mechanisms, development and spread ... 21
Surveillance of microbial antibiotic resistance, antibiotic consumption and nosocomial infections ... 24
Hospital hygiene and factors affecting nosocomial infection rates ... 24
AIMS ... 26
MATERIALS AND METHODS ... 28
Patients and settings ... 28
Susceptibility testing and bacterial isolates ... 30
Antibiotic consumption ... 30
Laboratory and physiological parameters ... 31
Questionnaire on ICU characteristics and Infection control ... 31
Genotyping of Pseudomonas aeruginosa ... 31
Detection of Metallo-β-lactamases in Pseudomonas aeruginosa ... 31
Antibiogram-based cluster analysis of Pseudomonas aeruginosa ... 32
Statistical methods ... 32
RESULTS ... 34
ICU characteristics and infection control ... 35
Antibiotic consumption and prescriptions ... 37
Bacterial species and antibiotic resistance ... 39
TABLE OF
CONTENTS
DISCUSSION ... 44
Main findings ... 45
Settings ... 45
Antibiotic consumption ... 46
Testing for bacterial antibiotic resistance and breakpoints ... 52
Bacterial isolates, antibacterial drug resistance and the emergence of resistance ... 54
Genotyping methods ... 57
Validation of antibiogram-based cluster analysis ... 58
Adherence to hospital hygiene procedures, hygiene factors and infection control measures ... 58
Validation of the ICU-STRAMA database ... 60
How to continue the battle against multiresistant microbes in Swedish ICUs ... 61
CONCLUSION ... 64
ACKNOWLEDGEMENTS ... 00
LIST OF
PUBLICATIONS
This thesis is based on the following original papers:
Erlandsson C-M, Hanberger H, Eliasson I, Hoffmann M, Isaksson B, Lindgren S, Nilsson L, Sörén L, Walther S. Sur-veillance of Antibiotic Resistance in ICUs in Southeastern Sweden. Acta Anaesthesiol Scand 1999; 43: 815-820
Walther SM, Erlandsson M, Burman LG, Cars O, Gill H, Hoffman M, Isaksson B, Kahlmeter G, Lindgren Sune, Nils-son LE, OlsNils-son-Liljequist B, Hanberger H & The Icu-strama Study Group (2002). Antibiotic prescription practices, con-sumption and bacterial resistance in a cross section of Swed-ish intensive care units. Acta Anaesthesiologica Scandinavica 2002; 46 (9): 1075-1081
Hanberger H, Erlandsson M, Burman LG, Cars O, Gill H, Lindgren S, Nilsson LE, Olsson-Liljequist B, Walther S and the ICU-STRAMA Study Group. High Antibiotic Suscepti-bility Among Bacterial Pathogens In Swedish ICUs. Scand J Infect Dis 2004; 36: 24-30
Erlandsson M, Burman LG, Cars O, Gill H, Nilsson LE, Walther S, Hanberger H and the ICU-STRAMA Study Group. Prescription of antibiotic agents in Swedish intensive care units is empiric and adequate.Scand J Infect Dis 2007; 39(1): 63-9
Erlandsson M, Gill H, Nilsson LE, Walther S, Giske C, Jonas D, Hanberger H and the ICU-STRAMA Study Group. Antibi-otic susceptibility patterns and clones of Pseudomonas aeruginosa from patients in Swedish ICUs. Submitted to Scand J Infect Dis
I
II
III
IV
V
AFLP Amplified Fragment Length Polymorphism APACHE Acute Physiology And Chronic Health Evaluation ARPAC Antibiotic Resistance Prevention And Control CDC Centers for Disease Control and Prevention (USA) CoNS Coagulase Negative Staphylococci
DDD Defined Daily Dosages
EARSS European Antimicrobial Resistance Surveillance System EDTA Ethylene Diamine Tetraacetic Acid
EUCAST European Committee for Antimicrobial Susceptibility Testing
ESBL Extended Spectrum Betalactamases ICU Intensive Care Unit
INSPEAR International Network for the Study and Prevention of Emerging Antimicrobial Resistance
ICNARC Intensive Care National Audit & Research Centre
HLGR High Level Gentamicin Resistant, refers to Enterococcus spp KISS Krankenhaus Infections Surveillance System
MBL Metallo-β-Lactamases
MIC Minimal Inhibitory Concentration MLST Multi Locus Sequence Typing
MRSA Methicillin Resistant Staphylococcus aureus MSSA Methicillin Susceptible Staphylococcus aureus MSSE Methicillin Susceptible Staphylococcus epidermidis NNIS National Nosocomial Infections Surveillance (USA) NPRS Nosocomial Resistance Prevalence Study
PFGE Pulsed Field Gel Electrophoresis ReAct Action on Antibiotic Resistance
SARI Surveillance of Antibiotic Use and Bacterial Resistance in German Intensive Care Units
SIR Swedish Intensive Care Registry
SMI Smittskyddsinstitutet (Swedish Institute for Infectious Disease Control)
SRGA Swedish Reference Group for Antibiotics
STRAMA Swedish Strategic Programme for the Rational Use of Antimicrobial Agents and Surveillance of Resistance TMP-SMX Trimethoprim-sulfamethoxazole
VAP Ventilator Associated Pneumonia VRE Vancomycin Resistant Enterococci
Throughout history, infections have been a major cause of human death. Ignaz Semmelweis, amongst others, discovered the importance of basal hygiene, good antiseptic technique and procedures in the prevention of bacterial spread and nosocomial in-fections. Eighty years ago, Alexander Fleming provided hope in the fight against infections when he discovered penicillin. Since then, more antibiot-ics have been discovered and invent-ed, but the struggle against germs re-mains a great challenge due to their dramatic ability to adapt rapidly to new environments. Today, there are no new classes of antibiotics in sight. Therefore it is becoming increasingly apparent that there is a vital need to gain control over bacterial resistance. Nosocomial infections remain a major cause of mortality and morbidity. The problem is most apparent in intensive care units (ICUs), which care for the most critically ill patients. Most ICU patients are compromised and vulner-able as a result of disease or severe trauma. According to a prevalence study1, one in ten people admitted to
hospital is given an antibiotic for in-fection. The risk of acquiring a noso-comial infection in a European ICU is approximately 20% according to a large surveillance study2. Ultimately,
the pronouncements of globally fund-ed conferences on the problem of in-creasing nosocomial infections and bacterial resistance matter little if we
do not find ways to prevent transmis-sion between patients and personnel, and develop local hygiene routines and antibiotic policies. This thesis is a holistic work on antimicrobial anti-biotic resistance, antianti-biotic consump-tion and hygiene in Swedish ICUs.
Nosocomial ICU infections
In this thesis the meaning of nosoco-mial infection is an infection acquired during hospital admission. Noso-comial infections are a problem for hospitals worldwide and for ICUs in particular. ICU patients usually suf-fer from underlying diseases and are immunocompromised, which makes them especially vulnerable. Most ICU-acquired infections are catheter-related in some way and are dealt with under each subheading and not as a separate entity.
Ventilator associated
pneumonia (VAP)
The most common infection in the ICU is ventilator associated pneumo-nia (VAP). Almost 70% of the patho-gens responsible for VAP are Gram negative. Of these, Pseudomonas
aeruginosa and Enterobacter spp
are the most common.
Staphylococ-cus aureus was responsible for 18%
of infections according to Fridkin and co-workers3.
Urinary tract infections
The second most common cause of nosocomial infection in the ICU is urinary tract infection (UTI). Almost all patients in the ICU have urinary catheters, and therefore the patho-gens associated with infection are
Escherichia coli (18.2%), Candida albicans (15.4%), Enterococcus spp
(14.1%) and P. aeruginosa (10.9%)3.
Bloodstream infections
The third most common nosocomial infection in the ICU setting is blood-stream infection, (BSI) and this is most often associated with intravascular devices. Gram positive bacteria are encountered in 64% and Gram nega-tive bacteria in 19.5%. Fungi account for 11% of pathogens responsible for nosocomial central venous catheter infections. The most common organ-isms are coagulase-negative staphylo-cocci (CoNS), S. aureus, enterostaphylo-cocci,
Enterobacter spp and Candida spp
according to National Nosocomial Infection Surveillance (NNIS) data3.
According to the SOAP study (The Sepsis Occurrence in Acutely Ill Pa-tients), which investigated the inci-dence of sepsis among 3 147 patients in 198 European ICUs during 14 days, the commonest origins were the lung (68%) and the abdomen (22%)4.
Surgical site infections
According to Fridkin, the microbial organisms associated with surgical site infections (SSI) in the ICU setting
may be related to the unique flora in the individual ICU. The problem of SSI is more common in wards outside the ICU3.
ICU pathogens
Staphylococcus species
Staphylococcus species are Gram
pos-itive and include coagulase pospos-itive S.
aureus, coagulase negative (CoNS) Staphylococcus saprophyticus and Staphylococcus epidermidis.
Staphylococcus aureus
S. aureus causes soft-tissue and skin
infections such as impetigo, follicu-litis, furuncles, carbuncles and hid-radenitis suppurativa. But they also cause pneumonias, sepsis, toxic shock syndrome and are common in late onset VAP. According to the 1995 EPIC study and the recently published SOAP study, S. aureus is the most common ICU-bacterium 2, 4.
Methi-cillin resistant S. aureus (MRSA) is a concern for all healthcare personnel. The options for treatment are vanco-mycin, rifampicin, daptomycin and tigecycline.
Coagulase Negative Staphylococci
S. epidermidis is the major pathogen
among CoNS. It is part of the normal skin flora. CoNS is the most common cause of bacteraemia in the ICU2, 4. S.
epidermidis can live for months on
ICU, and is therefore especially likely to cause catheter-related infections.
Enterococci
Enterococci are facultative anaerobic Gram positive bacteria, which are a natural part of our human intestinal microflora. There are almost 20 spe-cies of Enterococci, but it is mainly
Enterococcus faecalis and Entero-coccus faecium that are responsible
for infections in humans. Enterococci cause urinary tract infections, endo-carditis, surgical wound infections, intra-abdominal and pelvic infec-tions, and abscesses5. Most
vancomy-cin-resistant enterococci (VRE) are E.
faecium. There has been a shift over
the years among cultured enterococci from E. faecalis, which used to be the more common, to E. faecium.
Enterobacteriaceae
The enterobacteriaceae are Gram negative bacteria, which are a part of the normal human intestinal flora. Common bacteria are E. coli,
Kleb-siella spp and Enterobacter spp.
They cause urinary tract infections, intra-abdominal infections and pneu-monias. They are often resistant to first-line treatment such as amoxicil-lin. They are increasingly resistant to ESBLs, providing fewer treatment op-tions other than combination thera-pies or carbapenems6.
Pseudomonas aeruginosa
P. aeruginosa is a Gram negative rod.
Most clinical isolates produce pyocy-anin and pyoverdin, which are blue and green pigments7. The bacteria
have a characteristically sweet smell.
P. aeruginosa is an opportunistic
pathogen, both invasive and toxo-genic, and rarely causes disease in healthy individuals. The bacteria can survive for long periods in moist envi-ronments and on hospital equipment. It is an important pathogen in noso-comial infections, especially in im-munocompromised patients, causing respiratory tract infections, urinary tract infections, bacteraemias, and wound infections in burns patients. It also causes external otitis, folliculitis, and keratitis in contact lens wearers6.
P. aeruginosa is adaptive and
promis-cuous and easily develops antibiotic resistance. Single antibiotic treatment is therefore not the best option.
Stenotrophomonas maltophilia
S. maltophilia is an aerobic Gram
negative bacterium of low virulence. It tends to grow in moist environ-ments. It colonizes different solutions used in the hospital setting and may, via these solutions, penetrate and dif-fuse into wounds, mucosal-barriers and urine. S. maltophilia can cause lower respiratory tract infections and bacteraemia, but one has to bear in mind that S. maltophilia, due to its low virulence, rarely causes infection and therefore other sources of infec-tion must be excluded8.
Acinetobacter spp
The term Acinetobacter spp usually refers to Acinetobacter baumannii. This is an aerobic Gram negative bac-terium usually recovered from patients who are immunocompromised or have been subjected to prolonged hos-pital admission. A. baumannii tends to colonize aquatic environments e.g. hospital solutions, sputum, urine and respiratory secretions. It has low viru-lence and if it infects humans, it af-fects organs with high water content, e.g. urine, peritoneum, cerebrospinal fluid, the respiratory tract and burns. The bacterium is resistant to many antimicrobial agents, and therefore it presents a challenge for the treating physician9.
Candida spp
Candida spp are yeast-like fungi
with several virulence factors. There are more than 100 species but only a few are clinically relevant in hu-mans. Candida spp has the ability to adhere to other cells and surfaces and produces acid proteases. It has the ability to transform into hyphae-like forms. It tends to colonize and infect neonates, elderly patients and the immunocompromised, as well as patients with indwelling catheters. Candida causes a wide variety of in-fections ranging from skin and soft tissue, respiratory tract, gastrointesti-nal, genitourinary tract and systemic infections. Candida albicans is the most commonly isolated of Candida
spp and together with Candida
gla-brata makes up between 70-80% of
invasive isolates cultured in the USA.
C. glabrata has become more
impor-tant due to its greater resistance, es-pecially to azoles and amphotericin B, and it is therefore increasingly found, as are Candida krusei, Candida
trop-icalis, Candida lusitaniae, Candida parapsilosis, Candida guilliermondi
and Candida dubliniensis10.
Antimicrobial drugs
Beta-lactam antibiotics
This group consists of penicillins, ce-phalosporins, carbapenems and mono-bactams. They have the chemical struc-ture of the β-lactam ring in common.
Penicillins
These were the first in this group of anti-biotics to be discovered. They were gen-erally effective against Gram positive bacteria, but groups of penicillins that were effective against Gram negative bacteria were later discovered, and these proved to be potent broad-spectrum an-tibiotics, especially when combined with β-lactamase inhibitors.
Cephalosporins
This is a group of antibiotics with bactericidal effect. This is achieved by inhibition of peptidoglycan that is needed for cell wall synthesis. The first generation of cephalosporins is primarily effective against Gram
positive bacteria, but the later second and third generations are more effec-tive against Gram negaeffec-tive bacteria. Fourth generation cephalosporins are broad spectrum antibiotics with ac-tivity against both Gram negative and Gram positive bacteria.
Carbapenems
These have a chemical structure that makes them highly capable of with-standing β-lactamases. They have the broadest antibacterial spectrum of the β-lactam antibiotics. They are active against both Gram positive and Gram negative bacteria, but not to intracellular bacteria.
Monobactams
These are synthetic monocyclic β-lactam antibiotics, derived from a bac-terium. They are inactivated by some β-lactamases and by all extended spec-trum beta-lactamases (ESBLs). They are mainly used in P. aeruginosa infections, but they are also active against
Entero-bacter spp, Serratia spp, E. coli, Kleb-siella spp, Haemophilus spp, Proteus
spp and Citrobacter spp.
Fluoroquinolones
The quinolones in clinical use today have a fluoro group attached to their central ring system. Their bactericidal effect is due to inhibition of bacterial DNA-gy-rase and topoisomeDNA-gy-rase IV. Quinolones are often used to treat intracellular mi-crobes because they easily penetrate the cell wall. The kidneys, and to a lesser
extent the liver, are the main elimination pathways for quinolones. Ciprofloxacin and levofloxacin are the most commonly used fluoroquinolones in Swedish ICUs. Ciprofloxacin exerts its effect on Gram negative bacteria and therefore is an op-tion in e.g. upper urinary tract infecop-tions and exacerbations of chronic bronchitis. In Sweden, levofloxacin is used in atypi-cal pneumonias due to its effect on both aerobic Gram positive and Gram nega-tive bacteria, e.g. Mycoplasma
pneu-moniae, Legionella pneumophilia and Chlamydia spp.
Macrolides
Macrolides have a lactone ring to which deoxy sugars are attached. Their main effect is bacteriostatic but in high concentrations they can also be bactericidal. They exert their effect by binding reversibly to the ribosome in the bacteria, inhibiting protein synthesis. They are mainly eliminat-ed through the liver. Macrolides are mainly effective against Gram posi-tive bacteria but not Enterococcus spp. In Sweden they are mostly used for treatment of atypical pneumonias and when allergy to penicillin is sus-pected or established.
Oxazolidinones
Oxazolidinones are organic and con-tain a ring of 2-Oxazolidone with oxygen and nitrogen. Linezolid was the first antibiotic in this new class, and it exerts its effect by binding to a ribosome sub-unit, inhibiting
pro-tein synthesis in Gram positive bac-teria. The elimination of the drug is predominantly renal. It has a bacteri-cidal effect against most
Streptococ-cus spp and EnterococStreptococ-cus spp, and it
has a bacteriostatic effect against
Sta-phylococcus spp. Linezolid is mainly
prescribed for MRSA infections or other multiresistant bacteria as an alternative to the glycopeptide agent vancomycin.
Glycopeptides
Glycopeptides are non-ribosomal peptides consisting of glycosylated cyclic or polycyclic structures. Vanco-mycin and teicoplanin are two mem-bers of this group that are in clinical use. They inhibit cell wall synthesis by inhibiting the production of pep-tidoglycan. In ICUs vancomycin is used predominantly. They have a nar-row spectrum of action and are toxic to the kidneys and acoustic nerve. Plasma levels must therefore be moni-tored. Vancomycin is mainly used for severe multiresistant Gram positive infections, e.g. MRSA and MSSE. It is not absorbed when given orally but has a local effect on bacteria, includ-ing Clostridium difficile.
Aminoglycosides
Aminoglycosides are derived from Streptomyces or Micromonosporas. In the former case they are given suffix –mycin and in the latter –micin. They bind to a sub-unit of the ribosome and block initiation of protein synthesis.
They also makes mRNA misread, which also inhibits protein synthesis. In high doses they have dose-dependent nephro- and ototoxic effects, and therefore serum concentrations have to be monitored carefully. Aminoglycosides are used predominantly to treat infections with aerobic Gram negative bacteria such as
Enterobacter spp, P. aeruginosa and Acinetobacter spp.
Amphotericin B
Amphotericin B is derived from
Strep-tomyces nodosus and its name is
de-rived from its amphoteric properties. The mechanism of action is through association of amphotericin to fungal membrane ergosterols, which causes leakage of potassium and intracellu-lar components leading to cell death. Higher doses are fungicidal and lower doses are fungistatic. Amphotericin is used in the treatment of systemic fungal infections in immunocompro-mised patients. The agent is also ac-tive in candidiasis, aspergillosis, cryp-tococcal meningitis and visceral leish-maniasis. It is also used empirically in the treatment of fever in neutropenic patients that do not respond to broad-spectrum antibiotics.
Imidazole and triazole
derivates
Imidazole and the newer triazoles have a ring structure consisting of carbon, hydrogen and nitrogen. They exert their fungistatic effects through the inhibition of cytochrome 450 14-α-demythelase,
which is necessary for the conversion of lanosterol to ergosterol, which is used in the fungal cell walls. They are elimi-nated through the kidneys. Fluconazole is the most commonly used triazole. It is most effective against C. albicans and cryptococcal infections. A newer tria-zole, voriconatria-zole, is effective against
Aspergillus spp and all Candida spp.
Susceptibility breakpoints
In order to express antibiotic resist-ance, the terms susceptible (S), inter-mediate/indeterminate (I) and resistant (R) are used, which makes it easier for the treating physician to understand the resistance data obtained from cul-tures made at the microbiological lab-oratory. Several different systems are in use worldwide, and many countries have adopted their own. In Sweden, the Swedish Reference Group for An-tibiotics (SRGA)11 is responsible for
setting the MIC breakpoints as well as zone diameter breakpoints. Today in Europe, the European Committee for Antimicrobial Susceptibility Test-ing (EUCAST) is tryTest-ing to harmonise the MIC breakpoints between the EU countries12. Since the beginning
of 2007, all breakpoints in Sweden, except for macrolides and penicil-lins, correspond to EUCAST values. An inquiry is currently looking at the evaluation of the MIC values of the remaining antibiotics.
ATC, DDD and DDD
1000Antibiotic consumption was recorded using the Anatomical Therapeutic and Chemical Classification system (ATC) and Defined Daily Doses (DDD) that were developed during the 60s and 70s and adopted by WHO in 198213.
The system was invented and imple-mented for research on drug usage. The WHO Collaborating Centre for Drug Statistics Methodology classifies drugs according to the ATC-system and establishes DDD for each of these drugs. In order to compare data be-tween different countries and hospital settings, a preferred denominator has to be used. In the hospital setting the denominator most commonly used is 100 or 1000 patient days, giving the measure DDD/1000 patient days (DDD1000)13. The terms admission
days or occupied bed days are often used instead of patient days.
Antimicrobial drug
resistance mechanisms,
development and spread
Spread of bacterial
resistance
The use of antibiotics and antifungals drives the development of resistance in microbes, and several studies have demonstrated an association between increased antibiotic consumption and an increase in bacterial resistance to the drug in question14. The converse
relation has yet to be conclusively shown15, 16. In order to avoid
horizon-tal spread of resistance, it is of great importance that the prudent use of antibiotics is achieved and main-tained. Proper barrier precautions and isolation of patients must be used when indicated. Proper hygiene meas-ures must be undertaken, especially hand disinfection. Age and co-exist-ing diseases are important factors in the development of nosocomial infec-tions, as are length of stay and inva-sive catheters17, 18.
Resistance to
antimicrobial drugs
Resistance to antimicrobial drugs can be both acquired and intrinsic. The former is due to genetic mutations within the microbe resulting in better protection against the antimicrobial agent. The mechanisms behind this are multiple and complex but four main characteristics can be seen19. Firstly, the antimicrobial agent
may be inactivated, e.g. by the produc-tion of β-lactamases. Secondly, changes in accessibility may occur, by which an-tibiotics fail to enter the microbe. This happens when downregulation of porins takes place. Thirdly, antibacterial drugs may be excreted, e.g. when efflux pumps are upregulated. Fourthly, mutations can occur in the target for antibiotics render-ing the attackrender-ing antibiotic ineffective as it lacks a target. The production of alternative targets can shield the micro-organism or the target may be protected in other ways19.
β-lactam resistance
Resistance to β-lactam antibiotics is pro-duced by all of the above mechanisms.
Production of β-lactamases
As mentioned above, the production of enzymes that inactivate antibiotics is one mechanism of protection. This is the most common mechanism of resistance in Gram negative bacteria. β-lactamases inactivate penicillins, cephalosporins and to some extent carbapenems, by the hydrolysis of an amide bond in their β-lactam ring. The governing gene is often an integral part of plasmids and transposons, making them highly trans-ferable between bacteria20. β-lactamases
can be sub-grouped according to Am-bler classes A-D (AmpA-D)21. AmpB
β-lactamases are metallo-β-lactamases, and they have a broader hydrolytic ac-tion against all antibiotics in the β-lactam group22. AmpA β-lactamases are most
often inhibited by clavulanic acid, but inhibitor resistant enzymes like TEM and SHV are described. In contrast, AmpD β-lactamases are almost fully re-sistant to inhibition by clavulanic acid21.
Restricted- spectrum OXA-12 and ImiS are exceptions to this 23
,
as are theex-tended spectrum OXA-18 enzymes24.
AmpC β-lactamases are also of inter-est as extended spectrum β-lactamases (ESBL) as well as carbapenemases be-cause of their ability to disable most of the β-lactam antibiotics20.
Effect of porins and efflux pumps on intracellular concentrations of
β-lactam antibiotics
Porins in the outer membrane of the bacterium is a channel that some of the β-lactam antibiotics use to enter the microbe. Downregulation of porins or changes in their chemical structure pre-vents the antibiotic from exerting its ef-fects25.
Target alteration
The penicillin binding proteins (PBPs) are the targets of β-lactam antibiotics. There are several described types. If they are altered or downregulated, the effects of β-lactam agents will be abol-ished or reduced25.
Quinolone resistance
Multiple mechanisms are responsible for the development of quinolone re-sistance, but the result is a mutation in the genetic structure encoding for the DNA-gyrase termed topoisomerase, specifically, topoisomerase II and IV. In the latter, the mutation occurs in sub-units called gyrA and gyrB or in parC or parE19. The resistance mediated by
these changes can be enhanced by ef-flux pumps and porin permeability. Another mechanism involves the duction of a Qnr protein which pro-tects the topoisomerase from quinolo-nes. This is called plasmid-mediated quinolone resistance (PMQR)26. All of
the mechanisms can co-exist and have an additive effect on resistance levels.
Co-trimoxazole resistance
Co-trimoxazole (TMP-SMX) is a combination drug consisting of tri-methoprim and sulfamethoxazole. Both these drugs inhibit folate synthe-sis but at different stages. Resynthe-sistance occurs to both drugs. The most im-portant TMP resistance mechanism in Gram negative bacteria involves the alteration of dihydrofolate re-ductases (DHFR). This is encoded by
dfr-genes, which are integron-borne
genes. Resistance to SMX is mostly mediated by three sulphonamide genes, sul1-3, and they are transferred horizontally27, 28.
Macrolide and lincosamide
resistance
Resistance to macrolides and lincosa-mides are mediated by three different genes. The mefA encodes resistance to erythromycin. The ermB encodes resistance to both erythromycin and clindamycin, and ermA encodes an inducible resistance to clindamycin and resistance to macrolides29.
Aminoglycoside resistance
Several mechanisms are responsible for the development of aminoglyco-side resistance. Firstly, changes in cell permeability and decreased uptake which are chromosomally mediated30.
Secondly, mutations that produce al-terations of ribosomal binding sites can produce resistance31. Thirdly and
enzymes can produce high level re-sistance. More than 50 enzymes are known. The genes encoding for these enzymes are usually found in plas-mids and transposons32.
Surveillance of microbial
antibiotic resistance,
antibi-otic consumption and
noso-comial infections
There are several surveillance systems for bacterial resistance, antibiotic con-sumption and nosocomial infection rates. There is a strong focus on these issues as they represent major problems. The 58th World Health Assembly em-phasized this in 2005, when it stated that containment of antimicrobial resistance is a priority33.
Important information systems current-ly exist, including the Swedish Strate-gic Programme for the Rational Use of Antimicrobial Agents and Surveillance of Resistance (STRAMA), which was established in 199434. Germany has the
Surveillance of Antibiotic Use and Bac-terial Resistance in German Intensive Care Units (SARI)35 and its associated
German Hospital Infection Surveillance System (KISS). Denmark has its Dan-ish Integrated Antimicrobial Resistance Monitoring and Research Programme (DANMAP)36. In Europe, several
sys-tems co-exist: the European Antimi-crobial Resistance Surveillance System (EARRS)37, the European Committee
on Antimicrobial Susceptibility Testing (EUCAST)12, Action on Antibiotic
Re-sistance (ReAct)38 and Antibiotic
Resist-ance; Prevention and Control(ARPAC)39.
In the global arena, we have the surveil-lance system International Network for the Study and Prevention of Emerging Antimicrobial (INSPEAR)40.
Hospital hygiene and
fac-tors affecting nosocomial
infection rates
Several factors influence nosocomial infection rates, and the relations be-tween them are many and complex. The most important measures are the use of adequate barrier precau-tions, hand washing and isolation of carriers of multiresistant organisms. These aims can, however, be negated by heavy workload and a high staff turnover41.
Antibiotic policies affect bacterial drug resistance. It is very important that the correct antibiotic therapy is given42. The use of antibiotic cycling
has been studied and discussed and may have a role43, 44. Selective
de-contamination of the digestive tract has also been described, but it is not certain whether it has any effect on mortality45-47. The restrictive use of
The Centers for Disease Control and Prevention (CDC) in US has set out a campaign in 12 steps to prevent the spread and transmission of bacterial resistance. The guidance has four cor-nerstones: Prevent infection, diagnose and treat infection effectively, use an-timicrobials wisely and prevent trans-mission. A summary follows below51:
Prevent infection
Vaccinate patients against influenza and pneumococcal infection. Encour-age the vaccination of staff as well. Prevent conditions that can lead to infection, e.g. aspiration, pressure sores and dehydration. Remove un-necessary invasive devices and follow relevant guidelines when inserting them51.
Diagnose and treat infection
effectively
Use established criteria for infection, and target empiric and, when possi-ble, definitive treatment. Take appro-priate cultures. Use local resources e.g. specialists in infectious diseases when in doubt or complicated scenar-ios are encountered or foreseen, and know your local data51.
Use antimicrobials wisely
Use appropriate antibiotics and say no when there is no indication for treat-ment. Avoid long-term prophylaxis. Treat infections and not colonisation or contamination. Re-evaluate treat-ment constantly, and stop treattreat-ment
when infection has resolved or when infection cannot be proven51.
Prevent transmission
Isolate the pathogen. Break the chain of contagion and use barrier precau-tions. Perform hand hygiene, prefer-ably with alcoholic hand rub. Identify multiresistant organisms and take ap-propriate action51.
The general aim of the thesis was to investigate bacterial resistance and antibiotic consumption in Swedish ICUs.
Specific aims were the following:
– To try to correlate ICU demographic data with antibiotic con-sumption and antibiotic resistance.
– To try to help ICU physicians to interpret antibiotic resistance data so that they can prescribe the most appropriate antibacte-rial agents for the bacteria commonly found in the ICU. – To investigate on which clinical indications different
antibacte-rial agents are prescribed in the ICU.
– To investigate if and to what extent bedside physiological and laboratory data influence antibiotic prescribing patterns in the ICU.
– To investigate the emergence of resistance and transmission of Pseudomonas aeruginosa in the ICU using cluster analysis based on antibiograms and genotype data obtained by AFLP.
MATERIALS
AND METHODS
Patients and settings
In all papers, participating hospitals and ICUs are grouped into three cat-egories. They are labelled as i) tertiary care centres or university and regional hospitals, ii) district general hospitals, secondary hospitals or county hospi-tals and iii) local hospihospi-tals, primary hospitals or general hospitals.
Paper I
The first paper looks at ICU admis-sions in southeast Sweden. A total of eight ICUs were included, from five different hospitals. Three were dis-trict general hospitals (Norrköping, Jönköping and Kalmar), one was a local hospital (Eksjö), and one was a tertiary care university hospital (Linköping), which contributed pa-tients from the general, burns, cardi-othoracic and neurosurgery ICUs. A total of 17 592 patients were included. ICU demographic data were acquired including mean length of stay and to-tal number of admissions, and mean APACHE II scores were obtained from general ICUs.
Paper II, III,
The second and third papers include ICUs taking part in ICU-Strama. For paper II, 38 ICUs participated dur-ing 1999, coverdur-ing approximately six million of Sweden’s nine million in-habitants. For paper III, 29 ICUs par-ticipated during 1999-2000. ICU de-mographic data were studied in both
papers, although they were analysed in greater detail in paper II.
Paper IV
This study was conducted during the first two weeks of November 2000 and included 393 patients from 23 Swedish ICUs, of which 7 were terti-ary care centres, 11 district general hospitals and 5 local hospitals.
Paper V
Patients admitted to eight Swedish ICUs (five tertiary care academic hos-pitals located in Stockholm (Karolin-ska Huddinge and Karolin(Karolin-ska Solna), Gothenburg, Malmö and Linköping, and in three district general hospitals located in Stockholm (Södersjukhu-set), Jönköping and Skövde).
Paper V was based on material from the multi-centre Nosocomial Preva-lence Resistance Surveillance study (NPRS III) carried out in 2002, which investigated aerobic Gram negative bacteria cultured on clinical indica-tion. These hospitals were chosen to represent different geographical areas of Sweden. A total of 505 patients were included in NPRS III. Of these, 88 provided isolates of P. aeruginosa and were included in the study.
Susceptibility testing and
bacterial isolates
Paper I, II, III, IV and V
In all papers, bacterial samples were taken on clinical indications. Only initial isolates were considered in papers I-IV, whereas repeat isolates were also includ-ed in paper V. An isolate in this thesis is defined as bacteria cultured from a pa-tient admitted to an ICU. Susceptibility testing was done at the time of sampling by the disc diffusion (papers 1-4) and E-test (paper V) methods, as recommended by the Swedish Reference Group for An-tibiotics (SRGA)(accessed 23/7/2007)52.
SRGA-recommended breakpoints for susceptible (S), intermediate/indetermi-nate (I) and resistant (R) were used. Paper I considered the seven most com-mon bacteria cultured during the study period (Enterobacter spp, Klebsiella spp, Enterococcus spp, E. coli, Coagu-lase-negative staphylococci, S. aureus and P. aeruginosa.). A total of 800 Gram negative and 2 043 Gram positive iso-lates were collected.
Paper III specifically investigated
Aci-netobacter spp, CoNS, Enterobacter
spp, Enterococcus spp, E. coli,
Kleb-siella spp, P. aeruginosa, Serratia spp, S. aureus, and S. maltophilia. In order
to define which antibiotics were possible treatment options for each bacterium, a novel index was introduced called Treat-ment Alternative for more than 90% of tested bacteria (TA90).
In paper V, 101 P. aeruginosa isolates from 669 Gram negative isolates from the NPRS III were analysed. All samples were taken on clinical indication and they were cultured and tested at the local microbiological laboratory. In order to be able to detect the emergence of resist-ance, repeat isolates from each patient were also allowed. Five anti-pseudomo-nal drugs were investigated (imipenem, gentamicin, ceftazidime, ciprofloxacin, piperacillin-tazobactam). Isolates resist-ant or intermediately resistresist-ant to three or more β-lactam antibiotics were subjected to analysis for the production of metallo-β-lactamases with MBL Etest (AB Bio-disk, Solna, Sweden). We defined multi-drug resistance (MDR) as resistance to three or more of the tested drugs 53, 54.
SRGA breakpoints for MIC-values were used and accessed 07070411.
Antibiotic consumption
Paper I, II, III, IV
Data on antibiotic consumption using the Anatomical Therapeutic Chemical (ATC) classification system were provided by hospital pharmacies, expressed as antibiot-ics delivered in Defined Daily Dose (DDD) to the corresponding ICUs13. DDD is
calcu-lated as the average maintenance dose per day in adults for the main indication of the drug. In paper IV, administered antibiot-ic doses were recorded on a daily basis, in addition to the prescribing indication and the stop date or date for evaluation that were set on initiation of treatment.
Laboratory and
physiologi-cal parameters
Paper IV
Laboratory and physiological param-eters were recorded. These were body temperature, heart rate, blood pres-sure, breathing rate, urinary output, C-reactive protein, blood leucocyte count, platelet count, serum lactate, serum bilirubin, ALAT, arterial base excess, and arterial oxygen tension.
Questionnaire on ICU
char-acteristics and Infection
control
Paper I, II, III and IV
In papers I and III, each participat-ing ICU was asked to provide data on length of stay, number of admis-sions and severity of illness scores (APACHE II and III). In papers II and IV, more specific questionnaires were used to gather information on work-load and working procedures in each participating ICU, with questions on the utilisation of hand disinfectant, antibiotic treatment guidelines, regu-larity of rounds with specialists in in-fectious diseases, distances between ICU beds, and number of isolation rooms. Information was also gath-ered about how often feedback about antibiotic consumption was given by the local pharmacy and about local resistance patterns from the hospital microbiology laboratory.
Genotyping of
Pseu-domonas aeruginosa
Paper V
All bacterial isolates were investi-gated by amplified fragment length polymorphism PCR (AFLP). The method followed published protocols
55, except that EcoRI-0 primers used
for DNA amplification were fluores-cently labelled with Cy-5. PCR prod-ucts were detected by analysis of a 1-µl portion on an ALF Express DNA Sequencer (Amersham Pharmacia) as described previously 56. Similarity was
calculated by Dice coefficients using the BioNumerics uncertain band tool, with 0.5% tolerance and 0.5% opti-misation. Cluster analysis was done by the Unweighted Pair Group Method with Arithmetic Mean (UPGMA). All groupings with ≥ 90% similarity were inspected visually for the number of fragment differences. Isolates with ≥ 3 fragment differences were assigned to the same genotype.
Detection of
Metallo-β-lactamases in
Pseu-domonas aeruginosa
Paper V
Isolates positive for metallo-β-lactamases (MBL) with MBL Etest were subject to further analysis with imipenem +/- EDTA on Mueller Hinton agar. Isolates with a MIC-ratio imipenem/imipenem+EDTA ≥8 were subjected to multiplex real-time
PCR, targeting genes encoding the five groups of acquired MBLs, i.e. VIM, IMP, GIM, SPM and SIM57.
Antibiogram-based cluster
analysis of Pseudomonas
aeruginosa
Paper V
The 101 P. aeruginosa isolates from NPRS III were subjected to hierar-chical cluster analysis on the basis of log2 (MIC)-values of susceptibility to five tested antibiotics (imipenem, ceftazidime, piperacillin-tazobactam, ciprofloxacin and gentamicin). The analysis used the MiniTAB software package 58, 59. The distance measure
for each dimension was the absolute value of the difference between the log2 values, reduced by 1 (except for zero difference) taking the variability in MIC determination into account. The multidimensional measure used was Euclidian distance based on these unidimensional measures. Complete linkage clustering (farthest neighbour) was used, where the distance from a data point to a cluster is measured for the farthest data point in the cluster, and the distance between two clusters is measured using the most distant pair. This was done until the distance was zero and no more clusters could be combined.
Statistical methods
Papers I, II, III, IV and V
In paper I, statistical analysis was done with a non-parametric test (Pearson Chi 2 – test), and p-value was calcu-lated with Monte Carlo approxima-tion. In papers II and III, non-para-metric tests (Spearman’s rank correla-tion, Mann-Whitney, Kruskal-Wallis, Fisher’s test) were applied to explore relationships and differences60-64.
In paper V, adjusted Rand coefficient was used for overall concordance and the Wallace coefficient for directional information about the partition rela-tions65-67.
ICU characteristics and
in-fection control
Paper I
A total of 17 592 patients were includ-ed from January 1995 to December 1997. The annual number of patients treated decreased during the observa-tion period, and the number of admis-sion days decreased accordingly from 18 989 in 1995 to 16 850 in 1997. The annual mean length of stay, in days, ranged between 2 and 3.1 for general ICUs, between 1.7 and 1.9 for the cardiothoracic ICU, between 5.1 and 5.2 for the neurosurgery ICU, and between 13.9 and 15.5 for the burns unit. No correlation between APACHE II scores and antibiotic con-sumption was seen.
Paper II
Thirty-eight ICUs, providing primary services to a population of almost six million, participated in the study. Ten units were located at tertiary care centres (regional/university hospitals), 20 in secondary care centres (county hospitals) and eight were in local hospitals. The number of admissions and the total length of stay differed significantly between the ICU catego-ries (Table 1, page 36). Local hospi-tal ICUs had more admissions and shorter stays compared to county and regional hospital ICUs. 47% collect-ed illness severity scores but only one ICU computed mortality risk from these scores. The mean APACHE II
scores were slightly higher in regional hospital than in the local hospitals. 85% of ICUs had alcohol hand dis-infection available at the bedside, for more detailed information see Table 1. 44% had regular rounds with a spe-cialist in infectious diseases. This was more common in larger units (p=0.07), see Table 1.
Paper III
Twenty-six ICUs participated and 25 of these had alcohol hand disinfec-tion by each bed. More than 90% had isolation rooms available, 81% had a consultant in infectious diseases avail-able for rounds at least twice weekly and 62% registered severity of illness scores.
Paper IV
Twenty-three ICUs agreed to partici-pate. Seven were regional/university ICUs, 11 county, and seven local hos-pital ICUs. Out of a total of 393 pa-tients, 44% were admitted to regional ICUs, 43% to county ICUs and 12% to local hospital ICUs. 22% of the patients were already admitted (inpa-tients) at the start of the study.
Table 1
Characteristics*
Annual no. of admissions median (range) No. of beds median (range)
Mean APACHE II scores median (range) Mean length of stay (days) median (range)‡ Antibiotic consumption (DDD1000) median (range)
Written guideline on distance between beds Written guideline on the use of antibiotics Regular rounds with infectious disease specialist
Rounds with ID-specialist at least 5 days/week Hand disinfection, bedside §
Report on antibiotic usage at least once a year Report on antibiotic usage at least every 3 months
Report on bacterial species and drug resist-ance at least once a year
* Postoperative patients were included in some units,
leading to a large number of admissions and short mean lengths of stay. † P-values refer to comparisons between intensive care unit (ICU) categories. ‡ Correlated with total antibiotic usage (P=0.03, see text).
§ Negatively correlated with total antibiotic usage (P=0.05, see text). DDD1000, defi ned daily doses per 1000 occupied bed days. The number of units (n) varies as a result of missing values. Unless otherwise stated the ICU characteristics did not correlate with antibiotic consumption.
Local hospital ICU 2070 (1577–4955) n=5 8 (6–11) n=5 10.4 (10.0–12.8) n=3 1.0 (0.3–1.2) n=4 1072 (807–1377) n=4 0/5 (0%) 1/4 (25%) 4/5 (80%) 0/5 (0%) 4/5 (80%) 3/4 (75%) 2/4 (50%) 2/5 (40%)
County hospital ICU 1746 (591–4950) n=18 8.5 (6–19) n=20 12.0 (8.7–16.0) n=12 1.4 (0.6–3.2) n=18 1170 (604–2415) n=17 1/20 (5%) 4/20 (20%) 20/20 (100%) 9/20 (45%) 18/20 (90%) 16/18 (90%) 10/18 (56%) 3/17 (18%)
Regional hospital ICU 1042 (700–1490) n=9 9.5 (6–16) n=8 12.9 (12.7–13.0) n=2 2.3 (1.4–4.5) n=9 1541 (584–2247) n=9 2/8 (25%) 2/9 (22%) 9/9 (100%) 6/9 (67%) 7/9 (78%) 7/8 (88%) 5/8 (63%) 1/6 (17%) P-value† 0,03 0,53 0,36 0,01 0,18 0,19 1 0,15 0,07 0,51 0,76 1 0,68 Intensive care unit characteristics and selected practice parameters.
Figure 1
Change in antibiotic consumption in ICUs in southeast Sweden 1995-1997
6000 0 1000 2000 3000 4000 5000 1995 1996 1997 DDD Cephalosporins Isoxazolyl penicillins Quinolones Macrolides Carbapenems Penicillinase sensitive Pc Imidazoles AminoglycosidesLincosamides V ancomycin Broadspectrum Penicillins
Antibiotic consumption
and prescriptions
Paper I
Antibiotic consumption expressed as DDD decreased by 13.3%, but this figure fell to 2.5% when corrected for admission days (DDD/1 000 admis-sion days). Consumption of carbap-enems increased as the consumption of cephalosporins, macrolides and penicillins decreased (Figure 1). No correlation was found between sever-ity of illness scores (APACHE II) and antibiotic consumption.
Paper II
The median consumption of antibac-terial agents was 1 257 DDD/1 000 admission days. No correlation be-tween antibiotic consumption and se-verity of illness scores was observed. Antibiotic consumption was on aver-age 1.6 times higher in ICUs where no bedside alcohol hand disinfection was available. Total consumption of anti-biotics varied up to fourfold between the units but with no differences be-tween the ICU categories (Table 1). ICUs where a consultant in infectious diseases was responsible for antibiotic prescribing had lower consumption rates for glycopeptide antibiotics but for no other antibacterial agents. Lo-cal hospitals (primary hospitals) had significantly lower carbapenem con-sumption compared to university hos-pitals (tertiary hoshos-pitals) (Figure 2, page 38). Cephalosporins were the
most prescribed group of antibiot-ics (median 26%). ICUs with many admissions and short mean length of stay had lower median antibiotic consumption (p=0.01 and p=0.03 re-spectively). 21% of participating ICUs had written guidelines for antibiotic prescribing.
Paper III
Median antibiotic consumption was 1 391 DDD/1 000 occupied bed days in tertiary care centre ICUs, 1 201 in county hospitals and 983 in local hos-pitals. These differences were not sta-tistically significant (p=0.125). A wide range was seen, 605-2 143 DDD/1 000 occupied bed days. Cephalosporins were the most prescribed antibiotics with 26% of the median consumption, followed by oxacillins, carbapenems and quinolones with 13%, 10% and 8% respectively. Prescription patterns did not vary between the three ICU categories. Cefuroxime was by far the most used cephalosporin (79%), followed by cefotaxime (13%) and ceftazidime (4.5%).
Paper IV
The median rate of patients on antibi-otics was 74% but displaying a wide range of 25-93%. This was especially evident in county and local hospi-tals where the range was 35-93% (median 67%) and 24-80% (median 38%) respectively. The highest me-dian prescription rate was found in tertiary care centres with 84% (range
Figure 2
NAME OF ANTIBIOTIC
Type of hospital... DDD/1000 occupied bed days
GLYCOPEPTIDES Local ... 12,4 County ... 24,7 Regional ... 37,4 AMINOGLYCOSIDES Local ... 13,8 County ... 29,6 Regional ... 35,0 BETALACTAMASE SENSITIVE PENICILLINS Local ... 86,7 County ... 56,3 Regional ... 32,4 IMIDAZOLES Local ... 77,9 County ... 67,9 Regional ... 49,4 FLUOROQUINOLONES Local ... 85,0 County ... 88,8 Regional ... 106,1 CARBAPENEMS Local ... 58,1 County ... 116,5 Regional ... 165,9 ISOXAZOLYL PENICILLINS Local ... 168,0 County ... 162,7 Regional ... 277,4 CEPHALOSPORINS Local ... 314,2 County ... 330,2 Regional ... 366,1
Median consumption of antimicrobials in defi ned daily doses per 1000 occupied bed days (DDD1000) in different categories of intensive care unit. The consumption of carbapenems was signifi cantly lower in local ICUs compared with county and regional ICUs (P<0.05)
1996
100
50 150 200 250 300 350 400
0
58-87%). Almost half the patients re-ceived monotherapy, 20% had 2 anti-biotics and 3% had 3 and occasion-ally 4 antibiotics prescribed during the study period. Prior to admission, cefuroxime was the most commonly prescribed antimicrobial agent (mean 24%), but after admission carbap-enem was the most widely used. Van-comycin was rarely prescribed (2%). Linezolid and teicoplanin were not prescribed at all. Empirical therapy was the most common form of pre-scription (64%). No correlation was seen between laboratory parameters, such as CRP levels, leucocyte count and thrombocyte count, and antibi-otic prescription. Culture-based deci-sions were less common on days 1-2 than on days 3-14. A date for deciding whether to stop or continue antibiotic treatment was set in 8% of those re-ceiving empirical treatment and 3% of those who received culture-based therapy. 95% of antibiotics prescribed for sepsis were found to be appropri-ate when compared to antibiograms for blood isolates.
Bacterial species and
anti-biotic resistance
Paper I
A total of 2 043 Gram positive and 800 Gram negative isolates were taken on clinical indications. Only first isolates were considered. A significant increase in resistance among Enterococcus
spp was seen between 1996 and 1997 (p<0.001). This was due to a shift from
E. faecalis towards E. faecium. There
was a statistically significant increase in ciprofloxacin resistance among E. coli and Enterococcus spp (p<0.05). An out-break of methicillin-resistant S. aureus was seen in two hospitals during the study period, but no vancomycin resist-ance was seen in S. aureus or coagulase-negative staphylococci (CoNS). Resist-ance to oxacillin (≈70%), ciprofloxacin (≈50%), fusidic acid (≈50%) and netilm-icin (≈30%) in CoNS was seen through-out the study.
Paper II
All ICUs received preliminary infor-mation regarding bacterial growth in blood cultures. 74% also received this information for other specimens. More than half the units were given quarterly feedback on local levels of bacterial resistance. Almost 75% re-ceived this information at least annu-ally. Clinically significant levels of de-creased sensitivity to cephalosporins (second and third generation) were seen in Enterobacter spp and to ampi-cillin in Enterococcus spp. 26% of P.
aeruginosa isolates showed decreased
susceptibility (I+R) to imipenem, 11% to ceftazidime and 11% to cipro-floxacin.
Paper III
A total of 12 501 initial isolates were included. All were taken on clinical indications from patients admitted to participating ICUs during 1999-2000. The most common organism isolated was coagulase-negative sta-phylococci (CoNS) which constituted 17.5% of total isolates and 32.1% of blood isolates. This was followed by
Candida spp, E. coli and S. aureus.
The mean number of treatment
alter-Table 2
natives TA90, as described in material and methods, for E. faecium, CoNS,
P. aeruginosa and S. maltophilia was
1-2 per organism. Vancomycin was the only option for the first two and ceftazidime and netilmicin for P.
aer-uginosa. The treatment options for S. maltophilia were ceftazidime and
tri-methoprim-sulfamethoxazole. There were more treatment alternatives for the other bacteria, see Table 2.
Organism Acinetobacter spp Enterobacter spp E. coli Klebsiella spp P. aeruginosa Serratia spp S. maltophilia E. faecalis E. faecium CoNS S. aureus
a TA90 indicates an antibiotic to which > 90% of isolates of a given species or group of bacterial species are susceptible. b Numbers higher than 90 and thus defi ning TA90 are marked in bold. Ampicillin (AMP), cefotaxime (CTX), ceftazidime (CTZ), cefuroxime (CXM), ciprofl oxacin (CIP), clindamycin (CLI), fusidic acid (FUS), imipenem (IMI), netilmicin (NET), oxacillin (OXA), piperacillin-tazobactam (PTZ), rifampicin (RIF), trimethoprim-sulfamethoxazole (TSU) and vancomycin (VAN) c Including S (1%) and I (78%). According to the SRGA, wildtype E. coli are intermediately susceptible to AMP. d CTZ
The maximum numbers of isolates tested per antibiotic was 128 for Acinetobacter spp, 410 for Enterobacter spp, 778 for E.coli, 498 for Klebsiella spp, 602 for P. aeruginosa, 90 for Serratia spp, 198 for S. maltophilia, 805 for E. faecalis, 434 for E. faecium, 2238 for CoNS, 1063 for S. aureus.
TA90 (n) 3 4 7 6 2 3 2 3 1 1 6 CTX/ CTZ 21 67 99 97 92d 88 91d -CXM 6 41 91 83 -14 -CIP 89 93 94 95 85 94 68 1 0 4 -IMI 96 99 100 100 70 98 0 99 11 -NET 91 100 100 100 99 97 -0 0 54 100 PTZ 40 77 95 93 85 79 -TSU 96 93 92 95 -89 94 -41 -AMP -79c -100 22 -CLM -42 98 OXA -29 98 FUS -50 96 RIF -84 98 VAN -99 99 100 100
Treatment alternatives (TA90a) and susceptibility to antibiotics among tested pathogens. Proportion of susceptible isolatesb (%)
Paper IV
Among blood isolates (n=58) S.
au-reus, C. albicans and CoNS were the
most common organisms followed by
Enterobacter spp. In urine cultures
(n=38) E. coli (32%) and
Entero-bacter spp were the most prevalent
findings, followed by P. aeruginosa (7%), E. faecalis (7%) and Candida spp (4%). Respiratory tract isolates (n=44) had Klebsiella spp (18%), P.
aeruginosa (14%), CoNS (11%), S. aureus (11%) and H. influenzae (9%)
as the most common microbes. Empirical treatment was correct in 55/58 (95%) of bacteraemias accord-ing to correspondaccord-ing antibiograms. The instances of incorrect therapy included fluconazole for infection due to resistant C. albicans, meropenem for resistant CoNS, and cefuroxime for a naturally resistant E. faecium.
Paper V
The main source of isolates was the respiratory tract, from which 36
(35.6%) isolates were obtained. Of these, 15 (14.9%) came from the up-per airway (nasopharynx and tra-cheostoma) and 21 (20.8%) from the lower respiratory tract. Skin and soft tissue produced 25 (24.6%) isolates. Thirteen (12.9%) isolates were col-lected from abdominal wounds and drains. The urinary tract and blood contributed 12 (11.9%) and 10 (9.9%) isolates respectively, and five (5.0%) isolates were obtained from other body sites.
Six (5.9%) of the isolates were multidrug resistant (MDR), rising to eight (7.9%) when both intermediate and resistant isolates were consid-ered.
MIC distributions of all tested an-tibiotics are given in Table 3. No gen-tamicin-resistant strains were seen. Seventeen (16.8%) of investigated P.
aeruginosa isolates were resistant,
and four (4.0%) were intermediately susceptible to imipenem. Correspond-ing resistant and intermediate
num-Table 3
Imipenem Ceftazidime Piperacillin-Tazobactam Ciprofl oxacin Gentamycin ≤0,125 2 52 0,25 5 4 14 6 0,5 16 2 14 7 1 41 32 10 6 29 2 11 37 26 2 48 4 5 16 33 4 11 8 5 2 17 2 16 4 3 5 3 32 7 2 5 4 ≥64 5 3 5 Pseudomonas Aeruginosa Antimicrobial drugBold – Intermediate resistant (I)
Grey – Resistant (R)
a MIC-breakpoints according to SRGA and EUCAST 14/07/07 (www.srga.org, www.escmid.org) Minimal Inhibitory Concentration (MIC) mg/La
bers for ciprofloxacin were 10 (9.9%) and 6 (5.9%). For ceftazidime, eight (7.9%) resistant isolates were seen but no intermediates. The same was observed for piperacillin-tazobactam where 11 (10.9%) of the isolates were resistant.
Six isolates showed resistance to all investigated beta-lactam antibiot-ics and were subjected to phenotypic analysis for metallo-β-lactamases (MBL) with Etest. One isolate showed a MBL-phenotype, but no VIM, IMP, SIM, GIM or SPM genes were detect-ed with multiplex real-time PCR.
Eight patients with repeat isolates of
P. aeruginosa were found. In one
pa-tient P. aeruginosa was isolated from two samples taken from different body sites on the same day, and these isolates were therefore not considered as a true repeat isolate.
Genotyping of Pseudomonas aeruginosa
Amplified fragment length poly-morphism analysis (AFLP) of 101
P. aeruginosa isolates identified 68
genotypes. Fifty-one isolates were unique genotypes, and 17 genotypes displayed identical or similar patterns to one or several other isolates. Of these 17 genotypes, five were present in more than one ICU. Genotype A, C, H, M and N were present in 3, 2, 4, 3, and 2 hospitals respectively. We did not find any clonal spread of MDR clones in this study, but cross transmission between nine of 88 pa-tients (10.2%) was seen.
Antibiogram-based cluster analysis of Pseudomonas aeruginosa
The cluster analysis of the phenotypes based on MIC-values of P.
aerugino-sa for the key antibiotics investigated
